基于累积式自回归动平均传递函数模型的短期负荷预测

Short-Term Load Forecasting Based on ARIMA Transfer Function Model

针对短期负荷预测,提出了累积式自回归动平均(auto-regressive integrated moving average,ARIMA)传递函数模型的简化建模方法。传递函数模型考虑了干扰因素对因变量的作用,体现了干扰因素中变量间相互影响的关系。其构造灵活,可用较少的参数建立阶数较高的模型;并且假定值较少,容易得到满足。该文还将温度因素考虑在内,通过算例将传递函数模型和ARIMA模型的预测结果与实际值进行了比较,结果表明采用传递函数改进后的ARIMA模型预测精度提高,预测误差减小,具有较强的实用性。

The simplified modeling procedures of auto-regressive integrated moving average （ARIMA） model modified by transfer function （abbr. ARIMA transfer fimction model） for short-term load forecasting is presented. In this transfer function model the influence of disturbance factors on dependent variables is taken into account, thus the interaction among variables in disturbance factors is incarnated. Due to flexible structure of ARIMA transfer fimction model, a higher order model can be built with less parameters; there are less presuppositions in the modeling of transfer function and they are easy to satisfy. Considering temperature factor, the load of a certain province in spring is forecasted by univariate ARIMA model and ARIMA transfer function model respectively, comparison results of the forecasting results with actual value show that the forecasting result by ARIMA transfer function model possesses higher forecasting accuracy and the forecasting error is reduced, so the proposed method is practicable.